ESTRO 36 Abstract Book

S509 ESTRO 36 _______________________________________________________________________________________________

showed a 20% decrease in survival when exposed to Trastuzumab alone, but a combined treatment with radiation did not yield the expected decrease in survival, indicating an antagonistic interaction. Conclusion Our results show that before starting clinical trials, the combination of radiation therapy and combined targeting agents needs to be closely examined for each sub-type under consideration. The assumption that a combination of treatments will result in a synergistic response is clearly not always true. Acknowledgements We acknowledge funding from the Sydney Breast Cancer Foundation PO-0919 Stereotactic radiotherapy for brain metastases : Cyberknife versus VersaHD / ExacTrac M. Perdrieux 1 , M. Celeste 1 , I. Lecouillard 1 , E. Nouhaud 1 , C. Blay 1 , F. Jouyaux 1 , N. Delaby 1 , J. Bellec 1 , C. Lafond 1 1 Centre Eugène Marquis, Radiotherapy, Rennes CEDEX, France Purpose or Objective The aim of this study was to compare dosimetric and geometric performances of the CyberKnife (Accuray) and VersaHD (Elekta) with the ExacTrac system (BrainLab) in stereotactic radiotherapy for brain metastases. Material and Methods This study was conducted on 10 patients for Cyberknife M6 v10.6 with Iris collimator and VersaHD equipped with ExacTrac v6.1 and the Frameless system (BrainLab). The prescribed dose was 27 Gy in 3 fractions with 1mm margin between CTV and PTV for both modalities. The dosimetric study was also conducted with 2 mm margin for VersaHD plans in accordance to our clinical practices. Plans have been computed for CyberKnife with non- isocentric non-coplanar beams generated by inverse optimization on Multiplan v5.3 (Accuray) with the RayTracing dose calculation algorithm. For VersaHD, 4 non-coplanar arcs (VMAT) have been generated b y inverse optimization on Pinnacle v9.10 (Philips ) with the Adaptative Convolution algorithm. For each case, plans were normalized to obtain the same PTV co verage at +/- 0.2 %. The comparison was based on the brain volume outside PTV receiving 23.1 Gy. The volume of isodoses 6 Gy, 2.7 Gy and 1 Gy have been reported as well as the Paddick’s Gradient Index to characterize the dose gradient around PTV and the spread of low doses. Quality controls have been performed with Gafchromic EBT3 films (Ashland) and with an ionization chamber (Pinpoint 31014 /PTW) in an anthropomorphic phantom (STEEV/CIRS). The measured dose with film has been compared to the calculated dose according to the gamma index method with a 3% (local) / 2 mm criteria (analytical threshold : 30% of the maximum dose). The geometric shift between the measured and calculated dose distribution has been also reported. Results Table 1 shows that dosimetric criteria for plan validation were reached for both modalities and both margins. Compared to VersaHD, dose gradients obtained with Cyberknife were greater and lower volumes of healthy tissue received doses below 6 Gy. Ionization chamber measurements showed mean differences with the calculated dose of 2.53% and 0.03% for Cyberknife and VersaHD respectively. The mean value of the gamma index was 0.42 for the Cyberknife and 0.38 for the VersaHD. The mean geometric shifts between the measured and calculated dose distributions were 0.87 mm and 0.84 mm for Cyberknife and VersaHD respectively.

Conclusion For brain metastases stereotactic radiotherapy, Cyberknife with Iris collimator and VersaHD with ExacTrac both allowed compliance to dosimetric criteria. Cyberknife provided higher dose gradients than VersaHD and limited low dose irradiation of healthy tissues. The agreement between calculated dose and measured dose was acceptable for both modalities with mean gamma values lower than 0.5. An investigation will be performed to evaluate the use of low margins (1 mm) with the VersaHD / ExacTrac due to the very low geometric deviations. PO-0920 Utilizing monte carlo for log file-based delivery QA C. Stanhope 1 , D. Drake 1 , M. Alber 2 , M. Sohn 2 , J. Liang 1 , C. Habib 1 , D. Yan 1 1 Beaumont Health System, Radiation Oncology, Royal Oak MI, USA 2 Scientific RT, Munich, Germany Purpose or Objective The purpose of this study is to (1) investigate the feasibility of using Elekta’s R3.2 Log File (LF) Convertor as a standalone technique for patient-specific QA, and (2) assess Scientific RT’s SciMoCa monte carlo (MC) algorithm Eleven clinical, dual-arc VMAT patients [9 H&N, 2 low dose rate brain (35MU/min)] previously planned in Pinnacle and calculated using Adaptive Convolution (CS) were selected for this study. Arcs were delivered on Sun Nuclear’s ArcCHECK (AC) phantom and LF recorded. LF were converted into dicom plan files and calculated using CS and MC. For MC, all LF samples were reconstructed with no increase in calculation time. For CS, plans were reconstructed using 1° control point spacing to decrease computational cost. Original (Plan), LF, and AC doses were compared; statistical distributions (mean ± σ) of percent diode dose error, as well as 1%/1mm gamma pass rates, were calculated and compared for the five comparisons C1 to C5 shown in Table 1. A standard 10% threshold was utilized for both statistical and gamma analyses. Dosimetric degradation due to increased control point spacing (1/2/3/4°) was assessed for CS using 1%/1mm gamma criteria for 4 H&N and 1 brain patient. Delivering a 25x25 arc at various dose rates (35 to 570 MU/min) diode sensitivity dependence on dose rate was quantified. Results In-field diodes under-responded by 1.5±0.4% at 35 MU/min compared to 570 MU/min. Consequently, the four brain fields yielded lower Plan-MC pass rates (44±8%). These arcs were excluded from subsequent gamma analysis. Pass rates and diode dose errors are shown in Table 1. Comparing C2 to C1, MC and CS are compared. MC resulted in decreased σ values for 17/22 arcs (-3.7 ± 6.5%) and increased passing rates for 10/18 for use in said system. Material and Methods